jased 
The 
pased 
mage 
, two 
] data 
of the 
er to 
srtain 
d are 
ed to 
its. of 
aying 
y can 
.asily 
when 
lly, a 
cora 
n can 
ch of 
11 for 
ween 
vhich 
e tool 
»veral 
nation 
world. 
(e.g. 
riefly 
n out- 
nto a 
) is à 
| map 
nce in 
superimposition may both complicate visual analysis 
and either hide or distort spatial patterns that exist. 
Another deficiency is that the landscape boundaries are 
generally fuzzy, which means that they rarely conform 
to the distinct lines shown on a map. finally maps can 
deceive, or in other words, maps may contain 
inadequate or misleading information. 
2.2 Images 
Airborne and especially spaceborne images provide a 
synoptic view of the landscape and can be collected at a 
various temporal, spectral and spatial resolutions. The 
various forms of remotely sensed data, which can be 
acquired at microscopic and macroscopic levels, provide 
up-to-date unabridged information of the real world. 
They are, however, more complex to understand and 
handle than the vector data. 
In Living Proof, Petersen [1990] states that there are 
two data sources about an event: primary sources and 
secondary sources. Primary sources are actual 
eyewitnesses to an event and secondary sources are 
interpretations of the primary source. Images are 
primary sources. They are complete records of the 
sensor's view of a particular area, from a particular 
location, at a specific instance in time, and thus, are 
very useful especially if there is a question as to the 
reliability and currency of a secondary source. For 
example, any ambiguities associated with either the 
interpreted spatial location or the identification of 
geographical features may be resolved by querying the 
original source. Recent advances in computer aided 
technology have provided for the integration of 
remotely sensed imagery with maps, within a GIS, to 
overcome some of the deficiencies associated with 
using maps only as the primary source of information. 
The advantages of using remotely sensing information 
with other types of geographic data for spatial 
information management has been well documented. 
For example, Derenyi and Pollock [1990] discusses 
several ways in which remotely sensed data may be 
used to update existing maps in a GIS. Fung et al. 
[1993] outlines a system which allows the user to 
incorporate both satellite imagery and conventional GIS 
data sources for forest inventory management. Price 
[1995] and Wilson [1995] describe systems which use 
high resolution digital orthophotos to provide up-to- 
date, reliable geographic information for facilities 
management. 
Integrating remotely sensed data sources with digital 
maps in a GIS is not a trivial issue [Trotter, 1991]. 
Two factors which have impeded the integration process 
are that [Edwards, 1993] fundamental differences exist 
between the nature of the information resulting from 
digital image analysis and digital line maps; and the 
integration of image analysis technology with GIS has 
not progressed much beyond data exchange and 
simultaneous display. 
Images can play three distinct roles in a GIS: 
217 
1. A passive role, when the image serves as a backdrop 
to the various layers of information stored in vector 
form. It aids the viewer of such a composite to 
understand the geographic entities recorded as points, 
lines and polygons by portraying the information 
lost during data abstraction. 
2. A stand-alone role, when the image itself serves as a 
base map. The digital orthoimage maps, which are 
rapidly increasing in popularity, fall into this 
category. 
3. An active role, when images are incorporated in 
performing spatial analysis and data queries in a 
multi-layer geographic database. 
A scheme to facilitate the active role of images in a 
GIS is now presented. 
3. IMAGE SUPPORTED SPATIAL 
ANALYSIS 
Spatial analysis means to study the relationships 
between geographic features using spatial and non- 
spatial (attribute) data and to answer questions about the 
real world. 
A universal GIS (UGIS), one which can process and 
manipulate both vector based geographic data and raster 
based continuous tone image data, and is interfaced with 
a relational database management system (DBMS) 
provides the best environment for this purpose. 
Positional and attribute information can be accessed and 
cross-referenced quickly and efficiently. New thematic 
information can be generated easily or existing 
information updated and entered into the database. 
A usual procedure for a database query is to display the 
polygon layers of interest and initiate a search in the 
database by pointing on the entities whose attributes 
are sought or submitting a list of feature codes. The 
DBMS then returns the information requested. 
Two difficulties may occur in relying solely on a 
vectorized data representing geographic features: 
1. Displaying several data layers at once in 
superimposition complicates the location and 
selection of the entities visually. Therefore the 
queries are processed a few layers at a time. 
2. In the absence of suitable landmark features on the 
data layer queried or if the map layer is out of date, 
it may be difficult to find the correct geographic 
location of the information needed. This is 
especially true in exploratory investigations. 
The image based data query can alleviate these 
difficulties. The concept is rather simple. An image 
of suitable scale is displayed at the GIS workstation 
without any superimposition of vector data. An image 
provides full details of the land cover not only the 
outlines. It also contains qualitative information which 
can be assessed visually before the actual database query 
is issued, to localize the query area and thus reduce the 
processing time. The analyst then points with the 
cursor on the location in the image where information 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996